1 Introduction

Studies to understand the nature of seismicity spanning the last 60 years in the Koyna–Warna region have largely investigated the structure of the basement which hosts the earthquakes through determination of earthquake focal mechanisms, seismic velocities, density–susceptibility–conductivity changes, GPS velocities and other geophysical parameters (Talwani 1997; Gupta 2002, 2005; Gahalaut et al. 2004; Dixit et al. 2014; Gupta et al. 2016; Rao and Shashidhar 2016; Das and Mallik 2020). As a component of a megaproject on the reservoir triggered seismicity of Koyna–Warna, more recent geophysical information has been derived from borehole measurements, which also indicate the NW–SE direction as that of maximum horizontal principal stress (Arora et al. 2018; Goswami et al. 2020). Correlation of earthquake occurrences with changes in water levels of the Koyna reservoir has been conducted (Yadav et al. 2015, 2018; Gupta 2018 and references therein) and relations with pore pressure changes have been investigated (Chadha et al. 1997; Pandey and Chadha 2003; Hazarika et al. 2017; Yadav et al. 2017). Based on data from five continuous GPS sites, a northeastward anomalous motion of the eastern block is detected, causing loading on the faults, which sustain the seismic activity (Gahalaut et al. 2018).

The rugged topography and thick vegetation cover in the area have largely precluded studies of surface expressions of the active tectonics. Langston (1981) constructed a lineament map from LANDSAT images and showed a predominance of NNW and NNE trends defining an en echelon zone parallel to the Western Ghat Escarpment (WGE) in the Koyna–Warna region. Surface observations of vertical crustal deformation associated with filling and emptying cycles of the Koyna and Warna reservoirs were also found to be substantial from InSAR data. Das and Mallik (2020) have used SRTM data to calculate the morphometric parameters of the Koyna–Warna region and provided preliminary outlines of areas indicating high tectonic activity along the right flank of the Koyna reservoir and all around the Warna reservoir. An airborne LiDAR survey was carried out in April 2014 over an area of 1064 km2 covering southern part of the Koyna reservoir to south of the Warna reservoir, on the Deccan plateau. From the resultant sub-meter scale digital elevation model (DEM), Arora et al. (2017) extracted three major sets of lineaments in N–S, NW–SE and NNE–SSW directions and correlated these trends to subsurface fracture patterns observed in four boreholes surrounding the area as well as trends of subsurface structures based on magnetic anomalies (Arora et al. 2018).

With enhancements in seismological observations, clusters of small magnitude earthquakes continue to be observed along the northern part of the Donachiwada Fault, as well as clusters to the south of Warna reservoir (Shashidhar et al. 2020). In this work, we have investigated the DEM to trace the Donachiwada Fault on the surface. We have also analysed this particular feature in terms of its surface-subsurface expressions and association with major earthquakes.

2 Data and methodology

The LiDAR data was acquired on a Cessna 206 aircraft in the leaf-off month of April 2014, when vegetation cover is at its lowest so that penetration of the laser beam to the ground is maximised. The average flight height was 650 m above ground, flight speed around 160 km/hr. Laser pulse frequency was 125 KHz and scan frequency was 71. With a point density of 6.31/m2 on the ground, the final vertical accuracy of LiDAR data in the area is 10 cm and horizontal accuracy is 15 cm.

The ground GPS data is combined with the airborne GPS/INS data in the POSPAC software allowing the generation of a trajectory for the aircraft. The trajectory is then combined with the raw laser data in the Optech LMS software to generate the exact coordinates of the individual flight lines, which in turn provides the coordinates of the reflected points from the laser signals along each line. This is called a point cloud, which is examined in tiles of 1 km × 1 km. Based on the pattern of the terrain and vegetation type of the survey area, ground return pulses are identified. The orthophotos, point data, contours and cross-sections are used in combination for manual corrections of the point cloud of ground returns of the LiDAR laser pulses. Point-to-pixel conversion is achieved with the application of Tinning and Binning techniques. A DEM is generated using ground points only and a DSM using ground and unclassified points. Examples of DSM and DEM for one tile is shown in figure 1(a and b), respectively.

Figure 1
figure 1

(a) Digital Surface Model (DSM). (b) Digital Elevation Model (DEM) of a sample 1 km × 1 km tile based on airborne LiDAR data.

Full size image

The DEM over the KWSZ is used to trace the prominent surface linear features, which represent fault contacts. Their linkages with sub-surface trends as well as patterns of earthquake occurrence are analysed to derive certain conclusions.

3 Surface expressions of Donachiwada Fault

The LiDAR-derived DEM of the seismogenic region within the regional topographic map is shown in figure 2(B). The detailed morphology revealed by the metre scale DEM allows us to trace the lineaments, which cut across the rugged topography. The southern end of the Koyna reservoir is aligned NE–SW, from which the Koyna river emerges and turns abruptly eastward; the narrow valley of this river is the only significant E–W feature in the seismic region. The first turn of the Koyna river coincides with a prominent NE–SW lineament, which extends beyond the region of seismicity both to the east and west; we call it the Rasati Fault. Two distinct patterns of topographic features are notable: (i) NE–SW aligned ridges around the Koyna reservoir and to its south and (ii) NW–SE ridges flanking the Warna reservoir, namely the North and South Warna Faults, which continue westward into the Konkan. A location of field observation is indicated by the green triangle (1). The NE–SW ridges terminate against the northern flank of Warna reservoir, continue to the south on the southern bank of the reservoir. Overprinted on the above are subtle but persistent N–S trends running across the seismic zone from Koyna in the north to south of Warna, less than 10 km east of the Western Ghat Escarpment (WGE). The spatial pattern of epicentral locations of earthquakes of 2.0 ≤ M ≤ 5.1 for the period 2005–2018, is NNE–SSW south of Koyna, NW–SE along the northern Warna ridge and N–S across the Warna valley, along the edge of the WGE, closely following the topographic trends in the respective areas. The northern end of the Donachiwada Fault and the seismicity is truncated by the NE–SW lineament.

Figure 2
figure 2

(A) is the inset showing the location of Koyna within the Deccan Volcanic Province (DVP) in the map of India. (B) shows the DEM from airborne LiDAR (darker shaded) surrounded by topography from SRTM data; seismicity is marked by red stars, green line denotes the Western Ghat Escarpment (WGE); rivers and reservoirs are marked in blue. NW–SE and a prominent NE–SW topographic trends are shown in yellow dashes on the Deccan plateau as well as Konkan plain to the west, Donachiwada Fault inferred from seismology is indicated by a double dash yellow line. White rectangle demarcates the region of the N–S Koyna–Warna Fault, which is the area of focus in this study. Rectangles in black, south of Warna reservoir indicate fault plane orientation of two 2009 earthquakes, as deciphered from InSAR data. The green triangle (1) is the location of the cross-cutting fracture zone.

Full size image

The rectangle in white outline in figure 2(B) denotes the zone of study for the surface expressions of the N–S fault. The inferred alignment of the Donachiwada Fault (DF) from observations of helium anomalies and seismological data is marked by yellow dashed double line in the rectangle (Harpster et al. 1979; Talwani 1997; Shashidhar et al. 2017). Figure 3 shows the enlarged map of this area of study, covering the zone of the N–S lineament from SRTM 30 m DEM (left panel), LiDAR 1 m DEM (central panel), topographic sections along selected sections of the LiDAR DEM (right panel). West of the WGE and beyond the LiDAR area, the N–S topographic trends are more prominent and are readily visible in the SRTM data, sub-parallel to the Koyna reservoir and extending southward beyond the Warna reservoir. On the plateau the traces are subdued, only revealed from the new DEM. This is because the lava thickness is less than 500 m in the Konkan compared to about 1200 m on the plateau (Dhulipudi et al. 2019). From the topography along the profiles in the right panel, the surface locations of the N–S Koyna–Warna lineament can be traced in the form of narrow troughs, some deep, some quite shallow, appearing as notches on the ridges. Their continuity is persistent from north to south, with deeper depressions of up to 20 m in the north and only 7 m in the south. Such low amplitude features could be distinguished only because of the vertical accuracy of the LiDAR DEM. From the relative locations of the N–S lineament and the Donachiwada Fault, it is observed that the distances are within a few hundred metre in Koyna region with DF to the east, coincide exactly along P3 and P4 and separated by about 2000 m at P6 with DF to the west, in the Warna valley, the zone of strongest influence of the North and South Warna Faults. The N–S lineament continues to the south of Warna as in P7 and P8. We interpret the N–S lineament as the surface expression of the Donachiwada Fault, and name it the Koyna–Warna Fault, which traverses the entire region, extending beyond the epicentral zone.

Figure 3
figure 3

DEM of N–S Koyna Warna Fault from SRTM 30 m DEM (left panel), airborne LiDAR 1 m DEM (central panel), topographic sections P1–P8 across the Koyna–Warna Fault (right panel); red arrows on the right panel indicate the positions where the lineament is expressed on the surface topography and red crosses mark the location of the Donachiwada Fault on the same profile.

Full size image

Field photographs of the N–S and NW–SE lineaments at the locations marked as green triangles (2, 3, 4) are shown in figure 4(1–4). (1, marked in figure 2B) shows cross-cutting joints along the north bank of the NW–SE flowing Vashisti river, in the exact alignment as the South Warna Fault: the N14W fractures intersect the N64W fracture set and also dislocate them to the right, indicating sinistral movement. Rajaram et al. (2016) delineated the extensions of the NW–SE trends from Dharwar Craton to Deccan Volcanic Province and farther west. This suggests that the N–S ridges have developed later and modified the fabric of the NW–SE features. (2) is a long view of the Koyna–Warna Fault along its northern section, seen as a deep crevice; around the location of the Y junction south of the Koyna reservoir, it broadens into a zone of parallel fractures. (3) is a view of the Koyna–Warna Fault running N–S south of the Koyna river; the expression is that of a broader trough, becoming shallower as it meets the NW–SE ridge of the northern flanks of the Warna reservoir. (4) is a field photo of the N–S fault nearest to its shallowest and western most point; this geometry is due to the effect of the intersecting NW–SE North Warna Fault. Field photographs from the Warna area are not available due to entry restrictions into the Chandoli Tiger Reserve Forest.

Figure 4
figure 4

Field photographs (1) of fracture zone at location 1 in figure 2, showing N–S fractures having disturbed the NW–SE ones; (2, 3, 4) are field views of the N–S Fault, observed as continuous troughs on the ground.

Full size image

4 Sub-surface expressions of Donachiwada Fault

Tilt derivatives of magnetic anomalies from aeromagnetic data from flight heights of 1220 m and 2134 m (Agarwal et al. 2004) are shown in figure 5, left and right panels, respectively. The zero tilt (green band) represents the edge of the structures, brown colour represents positive tilts and blue represents negative tilt values (Verduzco et al. 2004). The left panel represents information till about 4 km depth, while the right panel has information till about 8 km depth, commensurate with flight heights. The shallower structures (left panel) have a NNE–SSW strike in the Koyna region, corresponding closely with the trends of the Koyna–Warna Fault/Donachiwada Fault. A prominent E–W trend along the Koyna river valley is notable, whereas the Rasati Fault has no signatures at this depth. In the Warna region NNW–SSE trends are seen, which become more NW–SE to the south. The North and South Warna Faults do not correspond to obvious structural trends here.

Figure 5
figure 5

(a) Tilt derivatives of aeromagnetic anomalies representing shallow subsurface structures, up to 4 km with epicenters of earthquakes M ≥ 5.0 and focal depths within 5 km, represented by red stars. (b) Tilt derivatives of aeromagnetic anomalies representing shallow subsurface structures, up to 8 km with epicenters of earthquakesM ≥ 5.0 and focal depths between 5 and 12 km represented by red stars. White dashed lines mark the Koyna–Warna Fault, the Rasati Fault and the North and South Warna Faults. The yellow dashed line represents the Donachiwada Fault.

Full size image

The right panel shows that in the Koyna region, the E–W trend of the Koyna river gets merged with a more extended NW–SE structure, while another E–W structural feature becomes prominent just north of the Warna valley. The Rasati Fault again does not correspond to any subsurface feature, except for a NE–SW turn of the Koyna subsurface feature, which is sub-parallel to it. Prominent NW–SE trends dominate the Warna region, extending westward beyond the WGE. The North and South Warna faults are aligned along them.

Earthquakes of M ≥ 5.0 have been plotted with their year of occurrence, magnitude and focal depth listed below the maps. Only three (3, 4, 8) have shallow focal depths (in left panel), while the others are deeper (right panel). Earthquakes in 1967(5), 1968(1), 1973(2) and 1994(12) occurred near and along the Donachiwada Fault, south of the Koyna reservoir; epicenters of (1) and (2) are also along the Rasati Fault. Earthquake in 1995(13) also occurred along the DF in the Warna valley. Earthquakes in 1980(3) and 1993(9,10) occurred along the Koyna–Warna Fault in the Warna valley region. Those in 1967(4), 1980(8), 1993(11), 1995(13), and 2000(16), occurred along the North Warna Fault, while in 1980(6,7), 2000(14,15) and 2005(17) the earthquakes occurred along the South Warna Fault. It may be noted that more than 60% of these earthquakes occurred after the impoundment of the Warna reservoir. Over the last six decades, since the beginning of triggered earthquakes in this region, the highest number of moderate magnitude earthquakes is located in the Warna valley, where the NNE–SSW and NW–SE faults intersect each other. A more recent earthquake catalogue records the larger number of earthquakes near Warna of much shallower focal depth (Shashidhar et al. 2019). Many seismological studies have reported different focal mechanisms for earthquakes in the northern and southern portions of the seismic zone: NNE–SSW or N–S strike-slip earthquakes in Koyna and NW–SW normal earthquakes in Warna (Rao and Shashidhar 2016). The two black rectangles south of Warna reservoir, marked on the map in figure 2(B), represent the modelled fault planes of two earthquakes of November and December 2009 of M = 4.7, 4.8, respectively (Arora et al. 2018), which show N–S and NW–SE orientations and normal faulting mechanisms (Rao and Shashidhar 2016).

We infer that the Koyna–Warna Fault is the Donachiwada Fault, which extends 10 km farther to the south. The cross-cutting N–S and NW–SE faults interact with each other in the Warna valley region, making this most prone to seismicity. While the seismogenic depths correspond to a relatively higher shear wave velocity, which can be attributed to higher rock strength (Rohilla et al. 2018), the local stress regime has a stronger effect at this junction, perhaps due to anomalous secular variation as reported by Gahalaut et al. (2018). High Vp and low to high Vp/Vs ratios are calculated for the source region from 3D local earthquake tomography, representing the heterogeneities in the shallow crust (Dixit et al. 2014; Kumar and Dixit 2017); the earthquakes occur in the regions of contrast of low to high Vp/Vs and the alignments of these zones follow those of the magnetic anomalies and surface trends.

5 Conclusions

This study highlights the efficacy of airborne LiDAR data for discovery of surface structural elements in rugged and vegetated terrains. In the Koyna–Warna region of reservoir triggered seismicity, analysis of the data reveals:

  • The high-resolution DEM from the airborne LiDAR survey has delineated the contrasts in geomorphological configuration of the northern Koyna region and the southern Warna region. These surface features have close match with subsurface features, inferred from seismology and aeromagnetic data.

  • The seismicity south of Koyna is located south of a NE–SW Rasati Fault, which is aligned along the first bend of the Koyna river as it turns to the east.

  • The Donachiwada Fault extends far into the Warna valley as the Koyna–Warna Fault.

  • The Warna region of cross-cutting Koyna–Warna Fault and North and South Warna Faults is where more than 60% of the M ≥ 5.0 earthquakes have occurred over the last six decades.